Part I.

The Diffraction Theory of Sea Waves and the Shelter Afforded by Breakwaters
Author(s): W. G. Penney and A. T. Price
Source: Philosophical Transactions of the Royal Society of London. Series A, Mathematical and
Physical Sciences, Vol. 244, No. 882 (Mar. 4, 1952), pp. 236-253
Published by: The Royal Society
Stable URL: http://www.jstor.org/stable/91516
Accessed: 18/11/2010 10:20

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=rsl.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

The Royal Society is collaborating with JSTOR to digitize, preserve and extend access to Philosophical
Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences.

http://www.jstor.org
 [ 236 ]

PART I. THE DIFFRACTION THEORY OF SEA WAVES AND

THE SHELTER AFFORDED BY BREAKWATERS

BY W. G. PENNEY, F.R.S. AND A. T. PRICE

CONTENTS
PAGE PAGE
1. Introduction 236 6. The transmission of waves through a
2. Fundamental equations 237 single gap in a breakwater 248
3. Diffraction of waves at the end of a long 7. The wave pattern and wave height be-
breakwater 238 hind the gap 250
4. Waves incident obliquely on the break--
water 243
8. Penetration of waves through a gap
smaller than one wave-length 253
5. The shelter afforded by an island break-
water 247 References 253

The diffraction of sea waves round the end of a long straight breakwater is investigated, use being
made of the solutions of mathematically analogous problems in the diffraction of light. The wave
patterns and wave heights are determined on both the leeward and windward sides of the break-
water, and for points quite close to the breakwater. This involves some extension of the calcula-
tions previously made for optical phenomena. The conditions obtaining in the lee of a small island
are discussed. The penetration of waves through a single gap in a long breakwater is examined,
and the result is shown to depend very much on whether the gap is small or not compared with
the length of the waves. The investigation was suggested by problems arising in the construction
of the Mulberry harbours.

1. INTRODUCTION

When sea waves impinge on a rigid obstacle, such as a breakwater, diffraction effects are
produced similar to those found when a beam of light is partly cut off by an opaque screen.
Consequently, when a reasonably exact picture is required of the shelter provided by a given
arrangement of breakwaters, these diffraction effects must be taken into account. The
mathematical theory of the diffraction of sea waves does not seem to have received much
attention,* in spite of the very considerable studies of surface waves in classical hydro-
dynamics, but fortunately this theory has many points of similarity with the well-known
theories of the diffraction of light waves and sound waves, and some of the mathematical
problems solved in those theories can be suitably modified and developed to give useful
information about sea waves.
We first show (? 2) that the general distribution of surface waves on a sheet of water of
any uniform depth, and their amplitude and phase at any point, are conveniently expressed
in terms of a complex function F(x, y), which we term the wave-function. The differential
equation and boundary conditions satisfied by F(x, y) are, in certain cases, identical with
those satisfied by a correspondingfunction in the theory of the diffraction of light. Hence by
making use of Sommerfeld's (1896) solution (see also Baker & Copson (I939)) for the dif-
fraction of light waves at the edge of a semi-infinite screen, we determine the wave-function
F(x, y) for the diffraction of sea waves round the end of a long, straight breakwater.The wave
* Diffraction of sea waves by obstacles of narrow breadth has been treated by Havelock (1940) in
discussing the wave resistance of ships; his results are not generally applicable to the problems we consider.
 PART I. DIFFRACTION OF SEA WAVES 237
pattern and the distribution of wave height on the leeward side are deduced both when the
incident waves are normal (? 3) and oblique (? 4) to the breakwater. These results are also
used (? 5) to obtain some idea of the conditions obtaining in the lee of a small island.
The mathematical theory of the diffraction of waves through a single gap in a breakwater
is developed in ?? 6 to 8; in this case, the general nature of the result is found to depend on
whether the gap is small or not compared with the length of the waves.
The assumption is made throughout part I that the height of the waves is small com-
pared with their length, and consequently the wave-profile is nearly sinusoidal, but it seems
probable that the results will not be altered seriously if the waves,are of moderate height.
The results will apply to water of any uniformdepth; if the depth is not uniform important
refraction effects may occur; for example, the waves will tend to turn so that their direction
of motion is along the line of greatest slope of the bottom, because the velocity of the waves
decreases with decreasing depth. Such effects would be important, and must be allowed
for, in water of moderate depth, but are not dealt with in the present paper.
With actual breakwaters the incident waves are partly reflected and partly destroyed
by turbulent motion. The present calculations deal only with ideal breakwaters in which
the waves are completely reflected. From a purely theoretical point of view it is convenient
to consider two types of ideal breakwaters, namely, (i) rigid breakwaters and (ii) flexible
or 'cushion breakwaters, as defined by equations (12a) and (12b) in ? 3. Of these only
(i) is probably of any practical importance, though it is of interest to remark that some
examples of (ii) were designed, constructed and tested during the war, and were found
to behave in the manner predicted by theoretical considerations.

2. FUNDAMENTAL EQUATIONS

Taking x- and y-axes in the plane of the undisturbed surface of the water, and the axis
of z vertically upwards, we denote by g the elevation at time t of the surface at the point
(x,y) above the undisturbed level. The motion of the water can be derived (Lamb 1924)
from a velocity potential 0, satisfying the equation

dX2
- =2
d- o dy2 +dZ2 = o(z).
(Z ?)

When the wave height is not too great, the pressurep is given approximately by
p =p(0-gz) (z<O), (2)
and in order that p shall be constant at the surface z =, we require, to the first order of
small quantities, =/g at z=. (3)

Also, since the normal component of the fluid velocity at the surface must equal the normal
component of the velocity of the surface itself, we have approximately

g=-z atz=O, (4)

whence, from (3) and (4)
~'+g = 0 at z=0. (5)
 238 W. G. PENNEY AND A. T. PRICE
Assuming the water to be of depth d, we have

0-=-0 atz=-d. (6)

Solutions of (1) which are periodic in t and satisfy the condition (6) are of the form
= A eikctcosh k(z + d) F(x, y), (7)
where F(x, y) is any complex function of (x,y) satisfying
d2F d2F
x2?- ky2 F (8)
and it is to be understood that the real part of the expression on the right of (7) is to be
taken.
From (5) we have -k2c2coshkd+gksinhkd 0,

so that c2 = tanh kd. (9)

The elevation C of the surface is obtained from (3) as

= Aikcekict
cosh kdF(x, y), (10)

where the real part of the expression on the right is to be taken. If F(x, y) is any complex
function satisfying (8), it is evident that the amplitude and phase of the surface wave will
be determined by the modulus and argument of F(x, y).
To represent progressive waves travelling, say, in the direction of the positive y-axis,
we may take the solution F(x,y) - e-ik (11)
Aikc
of equation (8), whence Ai= cosh kd,
eik(ct-y)
g
Akc
or taking the real part C= - coshkdsink(ct-y). (1 a)

This represents waves, of wave-length A = 2fr/k,period 27r/kcand amplitude

a = (Akccosh kd)/g,
travelling with the velocity c, given by equation (9).

3. DIFFRACTION OF WAVES AT THE END OF A LONG BREAKWATER

We assume that the breakwater extends along the x-axis from the origin to infinity. If
the breakwater is 'rigid', the normal component of the fluid velocity must be zero there,
so that d
= O atyO,x> O. (12a)

If the breakwater is of the 'cushion' type, the pressureremains constant and equal to the
hydrostatic pressure at all depths, so that from (2)

dt= 0 at y =0, x0. (12b)

 PART I. DIFFRACTION OF SEA WAVES 239
Although the cushion type of breakwater is probably of no importance from a practical
point of view (cf. ? 2), it is convenient to develop the theory of both types simultaneously
for reasons which appear later.
We consider first the diffraction of waves which are incident normally on the breakwater,
so that they are travelling in the direction of the y-axis, and are thereforegiven by equations
(11) and (lla).
To determine their diffraction, we have to find a solution of equation (8) which will
reduce to (11) when x is large and negative,and will satisfy (12a) or (12b) as the case may be,
at the breakwater.
In terms of F(x,y) the conditions (12a), (12b) become respectively
dF
O=- 0 at y = 0, x>0, (13a)
dy
and F=0 0, x>0.
aty= (13b)
The equations (8), (11), (13 a) and (13 b) are identical with the equations of Sommerfeld's
problem of the diffraction of light by a semi-infinite perfectly reflecting screen, the con-
ditions (13 a) and (13 ) correspondingto the cases where the light is (a) polarized in a plane
parallel to the edge of the screen, and (b) polarized in a plane perpendicular to the edge.
Sommerfeld's solution (Sommerfeld 1896; Baker & Copson I939) of these equations may
be reduced to the form

F(x,y)= 2 {e-ikY
fe- iui
du+- eiky e- i TiU2
d (14)

The + or - sign corresponds to the condition (13 a) or (13 b) respectively, and

a2 r24 '2= 4
(r-y), A= 2r7
(r+y), , (15)

where r V= (x2 +y2) and the signs for a and a' depend on the position of (x, y) in the four
quadrants as indicated below
y
(+'-) 0 (-,-)
x (16)
(+,) )(+, +)
Denoting by H the ratio of the maximum height of the waves at any point (x, y) to the
height of the incident waves, and the phase difference of the waves by e, we have
H =modF(x,y), (17)
e = argF(x,y). (18)
The expression (14) for F(x, y) can be evaluated by using tables of Fresnel's integrals,

cos2du and sin du,

or by graphical methods using Cornu' spiral (ahnke & Emde It convenient to

determine first the modulus and argument of the functionf(r), where

f(a) =(1 +i) -00O

e-Tiu2du. (19)

VOL. 244. A. 3I
 240 W. G. PENNEY AND A. T. PRICE
The graphs of modf(a) and argf(o) for values of between -3 and +3 are shown in
figure 1.
To obtain approximations tof(oa)for numerically large values of o, we find on integrating
by parts that, when a is negative,

/f( l(1+t) [--7exp - (1+f 2) +O(r-3)

=- 7 -2exp -- 1+2 2)} (20)

approximately, for large negative values of a. An approximation for large positive values
of a is then easily deduced from the relation
f(A) +f(- ) =1, (21)
which is true for all real values of o. We also find

f(a) 1> as a--+oo,

f(a) -> 0 as a>--oo, (22)
- at a =-0.
f(a) =
Using (22), we find the limiting forms of the solution (14) for various limiting positions
of the point (x, y) are as shown in (23) below.
F-> -e-iky

as y->+oo, x = 0
F~0
as x->+oo, y>0
F- eik lee of breakwater (23)
as x --oo for all y
F-> e-iky + i eiky
as y-> 0, x = 0
F-> e-iky + eiky
as x->+oo, y<0
F-> e-iky Z eikY
as y->--o, x= 0
Remembering that the elevation 5 is given by the real part of (10), and considering the
value ofF for large negative values of x, it is evident that this solution correspondsto waves
of the form (11 a) travelling in the direction Oy and incident normally on the breakwater.
On the leeward side of the breakwater F tends to zero for large positive x, agreeing with
the obvious fact that the screening becomes more and more effective as we move farther
I
away from the open end. At points directly behind the open end (x = 0), F tends to e-ik
for large y, indicating that the amplitude of the waves is there reduced to half that of the
waves in the open sea.
On the windward side of the breakwater, F contains the term eiky, corresponding to the
reflected waves moving in the negative y-direction. These combine with the incident waves
 PART I. DIFFRACTION OF SEA WAVES 241
to form stationary waves. At points well removed from the open end of the breakwater
(x large and positive), the amplitude of these stationary waves is twice that of the travelling
waves in the open sea. For a rigid breakwater, F-> eikY'+eieY as x->co for negative y, in-
dicating that the breakwater (y = 0) is at an antinode of the oscillating water surface,
i.e. the water-level varies by its maximum amount at the breakwater. On the other hand,
for a cushion-type breakwater, F->e-ikY -eiky as x->oo, showing that the breakwater is at
a node, so that the water-level remains constant.

n- / - / -0-8 4

2nf- / / -0?6 s,

317- / y c -0.4

FIGURE1. Graphs of the modulus and argument of the functionf(), where

arf(
=)(1 =(
f(cr) + i) eJdefru'du.du.
+i)f

41where (r ) (25)
theor- 3 sign beng taken according as x s positgative. 3
It follows from this
that2
to return to the general expression (14). On the lee side of the breakwater, for values of
sinceco< -4. t Hence
y> 2, the second integral in (14) will be theevalues
small ofr U we have, very(24), the
nearly,

F(x,y) =/(u) e-M2y/A, (24)

where c:FT/{4(r-y)/A}, (25)
When y is large comparedwitude and phase of the waves at intermediate positions we have
the - or + sign being taken according as x is positive or negative. It follows from this that
the conditions on the lee side of the breakwater at distances greater than about 2Aare nearly
r--y ( x,+y)y
the same whether the breakwater is of the rigid /2y
or cushion type. Using (17) and (24), the
approximately,(24)
maximum wave heights (H= modF(x,y)) have been determined as functions of x for
distances 2Aand 8t
distancesyy == 2t behind the breakwater.
8Abehind breakwater. These
These are shown
shown in figure
figure 2.
When y is large compared with x, we have

r-y == (x+y)2 -y - x2/2y approximately,

so that the value of r given by (25) is then approximately

- A y)
^( (26)
yFiu
31-2
 242 W. G. PENNEY AND A. T. PRICE

1.2-

O8- a
0.6-

6 5 4 '
3 2 1 0 -
1 ( ' I I l
r
\" O ,i
1
,
2
. I
3 "4
. .
5 .

1'-<s ^-^^

06- T\ b -
0.4 H
0.2- I

rigid breakwater
6 4 3 2 1 0 1 2 3 4 5 6
distance from end of breakwater measured in wave-lengths

FIGURE2. The maximum height of waves at normal incidence on a rigid breakwater, at distances
2A (b) and 8A (a) behind the breakwater. To obtain the wave heights at any distance y = nA
=
(where n > 4) behind the breakwater, multiply the values at y 8A by the factor J/(t-n).

8
FIGURE 3. Wave fronts and contour lines of maximum wave heights in the lee of
a rigid breakwater, the waves being incident normally.
 PART I. DIFFRACTION OF SEA WAVES 243
This shows that, for y large compared with x, the graph of the wave height as a function
of x has the same form for different values of y, but the scale for the x-axis is proportional
to /y. It follows that the graph of the wave height given in figure 2 for y .8Awill serve as
the graph of the wave height for any distance y- nAfrom the breakwater,where n is greater
than about 3, provided the numbers attached to the x-scale are multiplied by /(1n).
Another method, which is particularly convenient, for exhibiting the distribution.of
maximum wave height H behind the breakwater is to draw the contour lines
mod F(x, y) = H,
for a series of values of H. When y > 2/, the approximation (24) may be used, so that
mod F(x, y) = modf(or).
Hence the value (or values) of a, say -H, corresponding to any particular value of H may be
read from the graph of modf(oa)in figure 1. Equation (25) then shows that the corresponding
contour line will be the portion of the parabola

4(r-y) =- AO,

or in Cartesians, () + (2 (27)
for which x is positive when aH is negative, and x is negative when aH is positive. For points
nearer to the breakwater, the contour lines will differ appreciably from the above parabolas,
and their form will depend on whether the breakwater is of the rigid or cushion type. It is
then necessary to calculate them from the exact expression (14). The contour lines for a
rigid breakwater, corresponding to increments of 0.1 in H, are shown in figure 3. It will be
observed that there are narrow bands extending behind the breakwaterin the region beyond
the open end (x< 0) in which the wave height is slightly greater than in the open sea, but
in the lee of the breakwater the wave height is always less than one-half that in the open sea.
To obtain the wave pattern, i.e. the shape of the waveefronts, behind the breakwater,
we require the lines of constant phase. These are given by the curves arg F(x, y) = C, where
F(x, y) is given approximately by (24) for y > 2A. Beyond the open end of the breakwater
(x negative) o is positive, and it will be seen from the graph in figure 1 that argf(o) is then only
slightly different from zero. Hence the wave fronts are, as we should expect, only slightly
deformed from the straight-line fronts in the open sea. In the lee of the breakwater a is
negative and consequently argf(o-) is negative, which indicates a phase-lag. This phase-lag
increases with increasing x so that the wave fronts bend round towards the lee side of the
breakwater. The wave fronts are shown in figure 3. It will be observed that in the lee of the
breakwater, they are very nearly arcs of circles centred at the open end.

4. WAVES INCIDENT OBLIQUELY ON THE BREAKWATER

When the incident waves are travelling in a direction which makes an angle 00 with the
breakwater, it is convenient to divide the surrounding area into three regions Q, R and S
as in figure 4. The region S, where 0<0< 0, is in the main sheltered by the breakwater;
the region Q, where 0< 0< 2T, is relatively unaffected by the breakwater; and the region
R, where 27r- 00 < 0 < 27T, contains waves reflected by the breakwater.
 244 W. G. PENNEY AND A. T. PRICE
To obtain the wave heights and the wave pattern in these three regions, it is necessary to
generalize slightly the preceding solution. In place of the expression (14) for F(x,y), we
now have, using polar co-ordinates (r, 0) instead of (x, y),
F(r, 0) = exp{-ikrcos (0-00o)}f(a) ?exp{-ikrcos (0+00)} f(a'), (28)
wheref(o) is defined by (18), and
oa=2 (k)sin (0- 00), o' =-2(- )sin (+00), (29)
with 0 < < 27r.As before, the + or - sign is to be taken in (28) according as the breakwater
is of the rigid or flexible type.

/,S/

0Q O/0 . ?X___

4. Notation for the regions Q, R, S for oblique incidence of waves on a breakwater.

FIGURE
It will be observed that, in the region Q, a is positive and a' is negative; in the region R
both a and a' are positive; while in the region S both o and a' are negative. It will also be
seen that, if 00is put equal to iT-in (28) and (29), this solution reduces to the one previously
given for normal incidence.
By making use of (21), the expression (28) can be analyzed to show separately the
incident and reflected waves, and the diffraction effects. Thus in the region Q, since a is
positive there, we writef() = 1-f(- a) in (28), giving
F(r, 0) = exp{- ikrcos ( -00)} -exp{ ikrcos (O- 0)}f(- r)
? exp {- ikr cos ( + 00)}f('). (30)
The first term in this expression represents the incident waves, and the other two corre-
spond to the diffraction effects. The latter terms tend to zero as r tends to infinity, because
- a and a' are each negative and proportional to /r so that bothf( - a) andf(a') tend to zero.
Similarly in the region R, since both a and a' are positive, we apply (21) to bothf(r)
andf(o') in (28), giving
F(r, 0) = exp - ikrcos ( - 00)} exp- ikrcos (0 + 00)}
-exp{-ikrcos (0-00)}f(-o) exp{-ikrcos (0+00)}f(- '). (31)
In this expressionthe first two terms representrespectively the incident and reflected waves,
and the last two terms represent the diffraction effects.
In the region S both a and a' are negative so that only diffraction effects. as represented
by (28), are present.
 PART I. DIFFRACTION OF SEA WAVES 245

Thus the total effect of the breakwater can be regarded as the sum of (i) the shadow and
reflexion effects as given by geometrical optics and represented by the simple exponential
terms in (30) and (31), and (ii) the diffraction effects, which are given by the terms involving
the functionf(u) where u is negative in (28), (30) and (31).
An approximate expression for the diffraction effects for large r (except when 0 is nearly
equal to 00 or 27r--0) can be obtained by using the approximation (20). On substituting
this in equation (28) and using (29), we find for the region S

F(r,0) = 0-0796 /r) {cosec (-0) cosec (0o+ 0)}exp -2i( +l)}. (32)

For the regions Q and R, the last two terms in the expressions (30) and (31) lead to precisely
the same expression as on the left of (32). Hence, except when 0 is nearly equal to 00 or
27 -00, the diffraction effects are represented at sufficiently great distances by circular
waves diverging from the open end of the breakwater, as given by (32). The amplitude of
these 'diffraction' waves decreases like r-~ as r increases, except near the edge of the geo-
metric shadow (0 = 00), or near the edge of the region of reflexion (0 = 27T-0). .
Along the edge of the geometric shadow (0 = 00), the expression (28) for F(r, 0) reduces
to approximately I exp (- 2rir/A)for large r; this value is, in fact, a fair approximation for
values of r as small as 2A; for example, if 00 60?, we find
F(r,0) = 050 e-4i 0-07 e-(4'2)i for r 2A, (33)
the + or - sign corresponding, as before, to the breakwater being of the rigid or cushion
type. Hence along the edge of the sheltered region S the height of the resultant waves is
reduced to one-half that of the incident waves. Further inside S, each wave crest is bent
round approximately into an arc of a circle centred at the open end of the breakwater, and
its height is progressivelyreduced, in accordance with the expression (32), as the breakwater
is approached. The expression (32) also shows that at points well inside S there is a phase lag
of amount A/8. This indicates that, near the boundary of S, the wave crest is bent round
rather more than would be required to make it coincide with a circular arc centred at the
open end. These features are illustrated in figures 5 a, 5b and 6, which show the wave
patterns and indicate approximately the wave heights by the widths of the shaded areas.
On moving out of S into the open region Q, there is a discontinuity of one-half wave-length
in the phase of the diffractionwaves, which are represented in S by (28), and in Q by the last
two terms of (30) (cf. figure 5 b); the amplitude is, however, continuous and equal approxi-
mately to 0.5 at the boundary. This has the effect of making the resultantwave fronts and
wave heights continuous at the boundary, since near the boundary the diffraction waves
in the Q region subtract from the incident waves, thereby reducing their height to one-half
their normal height, and thus making them continuous with the waves inside the region S.
On moving farther into the Q region, the height of the diffraction waves progressively
decreasesand becomes zero at 0 = , i.e. along the continuation of the line of the breakwater.
In the case of the rigid breakwater, but not for the cushion type, there is a change of phase
again at 0 = ir of one-half wave-length, and as 0 is further increased, the height increases
again until it reaches the value 0*5 approximately at 0 = 2ir-00, i.e. at the boundary
between the Q and R regions. At this boundary there is once again a phase change of one-half
wave-length, the amplitude being continuous.
 246 W. G. PENNEY AND A. T. PRICE
The effect of the diffraction waves just inside the R region is to decrease the height of the
reflected waves near the R-Q boundary to approximately one-half their normal height.
These are superposed on the incident waves, so that the maximum height of the resultant
waves near the boundary is 1- times that of the travelling waves in the open sea.

geomnetric
shadow

/ breakwater-e

incident; K
waves waves
short-crested
5$
FIGURE 5b
FIGURE

1 short-crestedwaves.
FIGURE6
FIGURES 5a, b. Wave patterns for waves incident at 60? to a rigid breakwater. Figure 5a shows the
pattern according to geometrical optics. Figure 5b shows the diffraction waves.
FIGURE 6. The wave pattern for waves incident at 60? to a rigid breakwater. The pattern shown is
the superpositionof the geometricaloptics waves of figure 5a and the diffractionwaves of
figure5b. The width of the shadedareasrepresentingthe wavesgivesan idea of theirheight.
Inside the R region, the reflected waves combine with the incident waves to form a system
of short-crested waves, modulated both along and perpendicular to the breakwater, and
travelling parallel to it. This may be seen by multiplying the first two terms of (31) by
eikc, and taking the real part of the resulting expression. This gives for the combined waves

f/IO = 2 cos (y sin 00) cos (ct - x cos 00) (34)

if the breakwater is rigid, or
/C0 ==2 sin (y sin 0o) sin (ct-x cos 00) (35)
if it is flexible, where Cois the amplitude of the incident waves, and x and y are measured
parallel and perpendicular to the breakwater, with the origin at the open end. This shows
that in both cases we have a system of short-crested waves, of length 2n/cos 00 in the
x-direction and length 2fr/sin80 in the y-direction, travelling with velocity c sec 00 in the
x-direction. It will be observed that the wave velocity becomes infinite as 00-->jir (i.e. for
normal incidence), but at the same time the length of the crests in the x-direction becomes
 PART I. DIFFRACTION OF SEA WAVES 247

infinite, so that in the limit we have simply a system of stationary waves modulated in the
y-direction only.
These results are illustrated in figures 5 a, 5 b and 6, where the wave patterns are drawn
for 00 = 60?. Figure 5a shows the incident and reflected waves as given by geometrical
optics; figure 5b shows the diffraction waves and figure 6 shows the resultant waves. The
widths of the shaded areas representing the waves are a rough measure of their height.

5. THE SHELTER AFFORDED BY AN ISLAND BREAKWATER

A rough idea of the conditions in the lee of a small steep-sided island, when the waves are
coming from only one direction, may be obtained by treating it as a straight breakwater
of finite length, equal to the width of the island in the direction of the wave-front, this length
0-40

-30X
FIGURE7. The wave pattern produced by an
island of vertical sides, the waves being incident
normally, and the island being of length 10A.
030 The contours shown are those of areas in which
\ the wave height exceeds 0-15, 0.2, 0.25, 0.30
and 0.40 of the wave height in the open sea.
It is assumed that there is no correlation
/0925\ between the diffraction patterns from the two
ends of the island.

0?20

0-15

"-- Oh10<---
+f 1 + t. f1
waves of wave-length A

being a moderately large number (say 10 or more) of wave-lengths. The diffraction patterns
at the two ends will be mirror images of each other, but the irregularities in the shore of the
island and in the waves themselves will be sufficient to destroy correlation of these diffraction
patterns. We can, however, deduce that the maximum wave height at any point will not
exceed the sum of the maximum wave heights due to the diffraction round the two ends.
These maximum heights will be given approximately by the wave-height distribution shown
in figure 3 for one end, and by its mirror image for the other end. The distribution of wave
height behind an island breakwater of effective width 1OAhas been obtained in this way,
and is illustrated in figure 7.

VOL.244. A. 32
 248 W. G. PENNEY AND A. T. PRICE
6. THE TRANSMISSION OF WAVES THROUGH A SINGLE GAP IN A BREAKWIATER

We suppose now that there is a gap of breadth b, in a very long breakwater, on which
waves are incident normally. If the gap is small, i.e. b small compared with the wave-length A
of the incident waves, the gap will behave like a point source, semicircularwaves diverging
from it and having the same height practically all round any semicircle. If, however, the
gap is larger than a wave-length, the energy of the waves penetrating the gap will be propa-
gated mainly in the original direction of the incident waves; the wave pattern at consider-
able distances from the gap will still be approximately semicircles, but the wave heights will
vary along these semicircles, being greatest in the direction of propagation of the incident
waves. It will be shown that, when b is greater than A,a good approximation can be obtained
by superposition from the solution for the semi-infinite barrier already considered. In fact
this method gives an exactsolution for all values of b when one of the barriersis of the rigid
type and the other of the cushion type, and this solution is therefore given first, though it is
unlikely to be of practical interest on its own account.
We take the origin at the centre of the gap, the x-axis along the breakwaterand the y-axis
in the direction of propagation of the incident waves. The problem now reduces to finding
a function F(x, y), satisfying equation (8) together with the boundary condition dF/dy = 0
orF = Oat y = 0, Ix >I b, according as the breakwater is of the rigid or cushion type, and
containing a term e-iky for y negative, corresponding to the incident waves.
Now if we suppose for the moment that the part of the breakwater to the left of the gap
(i.e. for x < - -b) is removed, we have the case of a semi-infinite barrierfor which the solution
is given by (14). It has been seen that this solution may be analyzed into two parts, one of
which correspondsto the shadow and reflexion as given by geometrical optics, and the other
corresponds to diffraction waves diverging from the end of the barrier. In fact, on making
use of (19) and (21) as in ? 4 we can write the expression (14) for F(x, y) in the forms

F(x, y) = e-iky-f -g for x < lb and for all y, (36)

- ^+f?i-g 1 for x> -b and y>0, (37)
-e-iky ? e+iky -fiT g for x> b andy<0, (38)

where fi = (r, y) = e-ikyf(- (r y)), (39)

g = S(r, -y), (40)
and rI is the distance of the point (x, y) from the end of the barrier (sb, 0).
The diffraction waves in the region x <b, due to the barrier along the line segment
y 0=, x> ^b,are represented by the terms -fi gl (==F1 say) in the expression (36) for
F(x,y). From (39) and (40) we find that these diffraction waves satisfy the following con-
ditions along the unoccupied part of the x-axis:

Fl = , F=- 2d (rl,y) aty = O, x<b, (41)

for a rigid breakwater, and
F = 20(rl, 0), F 0 at y = O, x<b, (42)
for a flexible breakwater.
 PART I. DIFFRACTION OF SEA WAVES 249

Now if a second barrier is introduced along the line segment y =0, x < -b, the diffrac-
tion waves arising at the firstbarrierwould satisfyboundary conditions at the second barrier,
which are appropriate to a barrier of opposite type from the first. For example, if the first
barrier is rigid, the diffraction waves would satisfy the condition F1 0 at the second
barrier, which is the condition for a cushion barrier. Hence, if this second barrier is of the
cushion type, the diffraction waves from the first barrier will be unaffected by the presence
of the second barrier. It will, however, reflect the original incident waves and set up a
furtherset of diffraction waves which will satisfythe necessaryboundary condition dF/dy 0
at the first barrier.
Hence by superposing the diffraction waves for the two barriers we get the scheme of
values for F(x, y) shown in (43), which gives an exactsolution for the problem when one
barrier is of the rigid type and the other of the cushion type:

-fi +gl fi +gl

A2-g2 -f2 -g2
e-iky

j_J -fl +gl

cushion barrier f2 -g2 rigid barrier (43)
e-iky - e-iky e-iy + e-iky

-fl +g, -fl-gl

-f2 -+g2 --f2 --g2

It will be noted that this solution is exact for all values of b however small. The particular
case when b = 0 solves the interesting problem of a change from a rigid to a cushion reflector
at x = 0. There is a difference of phase of half-wave-length in the reflected waves in the two
regions, but the wave pattern is made continuous by thef and g terms; the wave height is
zero along the line x = 0.
When both barriersare of the same type, the boundary conditions at each barrier are not
automatically satisfied by the diffraction waves arising from the incident waves on the other
barrier, so that a secondary system of diffraction waves would need to be introduced to
obtain an exact solution. It will be shown, however, that these secondary waves are rela-
tively unimportant if b is greater than one wave-length. Thus, when both barriersare rigid
breakwaters, we consider the scheme of values for F(x, y) given in (44):

-fi +gl fi +gl

+f2 +g2 -f2 +g2
e-iky

rii -fl+gl
rigid breakwater -f2+g2 rigid breakwater (44)
e-iky + eiky e-iky + eiky

-fi +gl -fi -gi

-f2 -g2 -f2 +g2

32-2
 250 W. G. PENNEY AND A. T. PRICE
The boundary condition at either barriershould now be dF/dy = 0, but we find, for example,
at the left-hand barrier, that all the terms in Ffd/ycancel out, except the term dF1/dy,as
given by (41). Using (39) and (19), we find that
dF dF1 4?,i2_r_rf r 2\ /t
dy dy f
A-- 2A;) -(rl) expT2 A

and, to the degree of approximation given by (20), this is equal to

2i l/ i 2irir1 2 e(M 27rir
J/(2Arl) Aexp(4 (2Ar 4 0

Now (20) is correct to within 2 % of the maximum value of its modulus if a < -2; hence the
above condition is satisfied to this degree of approximation if r1>A. The minimum value
of r is the width b of the gap. We can therefore conclude that (44) is a good approximation
to the solution for two rigid breakwatersif the gap is not less than one wave-length in width.

7. THE WAVE PATTERN AND WAVE HEIGHT BEHIND THE GAP

The wave pattern will be given by the lines arg F(x, y) = constant and the wave height as
compared with that of the incident waves by mod F(x, y). The nature of the function F(x, y)
can be inferred from the equation (39) for 0(r,y), which determines the functionsfl, 2,
g1 and g2. The modulus and argument of 0 (r, y) can be obtained from the graphs in figure 1
for values of r?y up to 2*25A,a being negative, and for larger values we can use the approxi-
mation (20).
When r:y is zero, we have 0(r, ?y) = e-iky. Hence along the edge (x = ?b, y>0) of
one of the sheltered regions we havef1 = 2 e-ik, and along the edge (x = - b, y> 0) of the
other sheltered region we havef2 =1 e-iky.These are the largest valuesf1 andf2 can have,
for it will be seen from the graph off(a) that mod 0(r, ?y) is < 2 at all points, since the
correspondingvalue of ris always negative. Moreover, mod s$(r,?y) decreasesfairly quickly
at first as (-or) increases, i.e. as r?ty increases. For example, it is reduced to 0-08 when
ar= -2/2, i.e. when rTy = 2A. For larger values of rTy we have the approximate
expression ( +Y ( )
j/A(,
*v(yi
2,y}
exp 2iex
j87Ty)==/.2/
))exp * .(45)
8/N\2(r'y) } P1 4A
One immediate deduction is that, on the lee side of the breakwater (y > 0) the functions
g1 and g2 will be small except for points within one or two wave-lengths from either end of
the gap, and we may therefore ignore g, and g2 compared with the other terms in F(x, y).
This means that the wave pattern and the wave height behindthebreakwater are practically
the same for all the three cases, viz. both barriersrigid, both of the cushion type, or one of
each type.
At points opposite the gap ( -b <x < tb),we have

F(x, y) = eiky--f--
If the gap is broad,f, andf2 are small except at x = 5l 1b, where one of them is e-iky
' and the
other is small. This means, of course, that the incident waves pass through the gap almost
 PART I. DIFFRACTION OF SEA WAVES 251
unchanged except near tie edges of the gap, where their height is reduced by one-half.
When the gap is narrower,of the order of one or two wave-lengths, we can obtain an approxi-
mate expression for F(x, y) as follows.
Since r= y2 + (x- b)2, we have, when y is large compared with (x--b)

r--y -= (x- b)2/y approximately,

and similarly r2-y = -(x +b)2/y approximately.
Now from (39) and (19) we have
{f-w }
y(r,y)== ie-ik (1 +i) (1 -i) + e- riu2du ,

where w = /{4(r-y)/l}, and for small values of w we can expand the integrand as a power
series, giving
0(ry) = e-iky 1--w- .+i(-W+ .. .. (46)
Hence, using (46), we find
F(x, y) =e-eky-f i-f2

- e-iky b) eii + r (3x2 + b2) e-7ti} (47)

AAhy)1^^ ^^ }- (47)
This gives
modF(x,y) = (y) + 18 (3x2+b2)}, (48)

argF(x,y) = tan-( 3A t(3x 2 * (49)

3Ay+( (3x2 +:I b2)))-

The above approximation holds good for points behind the gap at distances large com-
pared with ?b-I x 1. The same approximation is found for points just inside the sheltered
regions, on using F(x, y) -=fi - f2 and taking Ix i - -b small compared with y.
The expression (49) shows that argF(x, y) lies between -ky and Tr-ky for the whole
range of values of x considered. Hence the waves in the region - b<x < b remain almost
parallel to the original incident waves and penetrate to some extent into the sheltered
regions. But at points well within the shelter of either barrier, the functionsf, andf2 may be
calculated from the expression (45) for 0(r, ?y), which shows that the waves eventually
curve round into a circular form, the waves corresponding to the separate termsfi andf2
diverging from the ends x = lb and x = -b of the gap.
The expression (48) for modF(x,y) shows that the waves are nearly of uniform height
b//(Ay) in the range -- b<x< b, though there is a slight minimum at x = 0. But it will
be noted that the second term on the right of (48) is very small for all admissible values of x,
so that the increase in height on moving outwards from x = 0 is also small. This expression
for the height will hold for a short distance into the sheltered regions, but at points farther
in, the functionsfl andf2 rapidly decrease, so that the height of the waves correspondingly
decreases as the waves curve inwards towards the breakwater.
The heights of the waves and their phase differences at various points for the case when
b = 2-5Aare shown in table 1.
 252 W. G. PENNEY AND A. T. PRICE
TABLE 1. WAVE HEIGHTS AND PHASE DIFFERENCES AT POINTS BEHIND
A BREAKWATER WITH A GAP OF WIDTH 25/A
X= distance from line through mid-point of gap in wave-lengths.
Y=distance behind gap measured in wave-lengths.
I X 0 0-5 1.0 1-25 1-5 2-0 4 8 12 16 20 24

5 1.01 0-92 0*73 0-57 0-42 0*29 0-21 0-04

280 24? 2? -13? - 30 -180? -304? -1975?
10 0-77 0-76 0-69 0*66 0*62 0-52 0.10 0-13 0-03
36? 330 22? 12? -1? -26? -185? -798?
25 0-51 0-50 0-49 0-47 0-44 0*43 0-39 009 0-05
41? 400 40? 23? 25? 10? -60? -405?
50 0-35 0*35 0-35 0*35 0-32 0-31 0*34 0-28 0'19 0.05 0.01
430 42? 40? 370 350 20? -10? -140? -328?
100 0-25 0*25 0-25 0-26 0-25 0-24 0*23 0*20 0-23 0-20 0-16 010
440 42? 400 390 390 370 23? -60? -207? -415? -675? -1040?

100X

100I

80S

60A -

1.10x-

20o

width ofgap 25X

I I I I I I . I

30X 20A 10 0 10 20 30A

FIGURE 8. Wave heights behind a gap of width 2*5A in a rigid breakwater,
the waves being incident normally.

The phase angles given in table 1 show the difference in phase between the incident waves
at the point (if they had not been disturbed by the breakwater) and the actual waves at the
point. The increasing negative value of these phase differences for increasing x indicates
that the waves are bent backwards behind the breakwater, a phase difference of 360?
corresponding, of course, to one wave-length. The wave pattern and distribution of wave
height behind the breakwater are shown in figure 8.
 PART I. DIFFRACTION OF SEA WAVES 253
8. PENETRATION OF WAVES THROUGH A GAP SMALLER THAN ONE WAVE-LENGTH

When b is small compared with A,an approximate solution may be obtained by assuming
that the stream-lines for the motion of the water through the gap are the same as for a simple
uniformstreaming of water through the gap. The mathematical problem then becomes
identical with the problem of the penetration of sound waves through a narrow slit; this
problem is treated by Lamb (1924). Using Lamb's solution we deduce the following
expression for F(x, y) for points not too near the gap:
(A/4r) exp (-i- -kri)
-
log (rb/4A)+y+ri n ' (
where r is the distance from the centre of the gap, and y is Euler's constant, equal to 0 5772.
This corresponds to waves diverging uniformly in all directions behind the breakwater
from the gap as a point source, the heights of the wave at any point being given by mod F,
which is proportional to r-.
The expression for the height of the waves (mod F) may be written in the form

mod F (51)
2= (kb{log(+kb+ y)2+ })I
and the coefficient of V(b/7rr) in this expression is slightly greater than unity when b lies
between 0.lA and 0-2A.Thus when b = A/2i7the value of the coefficient is 1-053. Hence for
this range of values of b, the height of the waves is given approximately by J(b/irr). This
may be compared with the approximate formula b/J(Ay) for the maximum height when
b>A. These two expressions would be equal at r = y if b2/Ay= bl/rr, i.e. b = A/fr,which
suggests that the formula b/J(Ay) for the maximum wave height gives a rough approxi-
mation for values of b down to about A/Tf.The fact that the maximum height is proportional
to b and not jb when the gap is not very small means that the energy of the wave motion
passing through the gap is still propagated mainly in the direction of the incident waves.

REFERENCES

Baker, B. B. & Copson, E. T. I939 Huygens'principle.Oxford.

Havelock, T. H. 1940 Proc.Roy. Soc. A, 175, 409.
Jahnke,E. & Emde, F. 1945 Funktionentafeln.
New York.
Lamb, H. I924 Hydrodynamics.Cambridge University Press.
Sommerfeld, A. I896 Math. Ann.47, 317.